04.1 - Aerospace Grade Materials, Structural Analysis, Fatigue and Damage Tolerance
TRAIN ON SYNTHETIC - TEST ON REAL: DOMAIN ADAPTATION FOR STRAIN-BASED DAMAGE DETECTION ON AN AIRCRAFT WING
P. Conen¹, F. Raddatz¹, G. Wende¹; ¹DLR, Germany
This paper evaluates the transferability of a machine learning model that is trained with synthetic data. The virtually trained model should predict the existence, location, and geometrical characteristics of cracks in the physical domain. This paper contributes to the fields of structural health monitoring and domain adaptation.